Recent Experiences and Future Projects with Big Data with the Community Earth System Model

Date and Time: 
2017 April 25th @ 3pm
ML main seminar
Gary Strand

The Community Earth System Model (CESM) is a state-of-the-art climate model and is NCAR’s flagship climate model. It is used to simulate the Earth’s climate system, from the distant past into the future, and also to investigate the processes underlying the climate system. Its components consist of an atmospheric model, ocean model, sea ice model, land surface model, smaller additional components, that are all interconnected via a “flux coupler”. The entire codebase is approximately 1.5 million lines, mostly in Fortran, and is typically executed on Top500-class HPC resources, including Yellowstone and the new Cheyenne system located at the NCAR-Wyoming Supercomputer Center (NWSC). A release of the latest version, CESM2.0, is expected in June 2017. Simulations done with CESM, depending on the specific model configuration, can replicate time periods from as short as few days to tens of thousands of years. Large ensembles of model runs are also created, in which the identical configuration is executed from dozens to thousands of times. As a result, large volumes of model data can be generated, from tens of terabytes to over a petabyte from a single project. The processing, management, and analysis of these “big data” projects can be a daunting task. New tools have been developed and are under development to assist in dealing with the scale of these projects. My talk will focus on recent CESM projects, including the Large Ensemble (LENS), Last Millennium Ensemble (LME), and the Decadal Prediction project (DP) that have employed these new tools. The upcoming Coupled Climate Model Intercomparison Project phase 6 (CMIP6) will also be discussed.

Speaker Description: 

Gary Strand is a software engineer in the Climate Change Prediction group of the Climate and Global Dynamics Division of NCAR. He began work at NCAR in 1986 as a student assistant, and has been involved in several generations of climate model development in CGD. He is the primary data manager and data scientist for the latest NCAR climate model, the Community Earth System Model (CESM). He has led the major data management activities and projects for the  CESM since 2003, including CMIP3 and the current CMIP5. He is also one of the key personnel for the Earth System Grid (ESG) project, participating since its inception in 2001. Gary has also created a number of visualizations of CESM output that have been used in many scientific presentations as well as in major broadcast media.

Event Category: